These last words allude to the various particular etymologies which had been enumerated by Sokrates as illustrations of the Herakleitean theory. They confirm the opinion above expressed, that Plato intended his etymologies seriously, not as mockery or caricature. That Plato should have intended them as caricatures of Protagoras and Prodikus, and yet that he should introduce Kratylus as welcoming them in support of his argument, is a much greater absurdity than the supposition that Plato mistook them for admissible guesses.
[65] Plato, Krat. c. 111, pp. 435-436.
Sokrates goes still farther towards retracting it,
These consequences are fairly deduced by Kratylus from the hypothesis, of the natural rectitude of names, as laid down in the beginning of the dialogue, by Sokrates: who had expressly affirmed (in his anti-Protagorean opening of the dialogue) that unless the process of naming was performed according to the peremptory dictates of nature and by one of the few privileged name-givers, it would be a failure and would accomplish nothing;[66] in other words, that a non-natural name would be no name at all. Accordingly, in replying to Kratylus, Sokrates goes yet farther in retracting his own previous reasoning at the beginning of the dialogue — though still without openly professing to do so. He proposes a compromise.[67] He withdraws the pretensions of his theory, as peremptory or exclusive; he acknowledges the theory of Hermogenes as true, and valid in conjunction with it. He admits that non-natural names also, significant only by convention, are available as a make-shift — and that such names are in frequent use. Still however he contends, that natural names, significant by likeness, are the best, so far as they can be obtained: but inasmuch as that principle will not afford sufficiently extensive holding-ground, recourse must be had by way of supplement to the less perfect rectitude (of names) presented by customary or conventional significance.[68]
[66] Plato, Kratyl. p. 387 C. ἐὰν δὲ μή, ἐξαμαρτήσεταί τε καὶ οὐδὲν ποιήσει. Compare p. 389 A.
[67] Plato, Kratyl. p. 430 A. φέρε δή, ἐάν πῃ διαλλαχθῶμεν, ὦ Κράτυλε, &c.
[68] Plato, Krat. p. 435 C. ἐμοὶ μὲν οὖν καὶ αὐτῷ ἀρέσκει μὲν κατὰ τὸ δυνατὸν ὅμοια εἶναι τὰ ὀνόματα τοῖς πράγμασιν· ἀλλὰ μὴ ὡς ἀληθῶς γλισχρὰ ᾖ ἡ ὀλκὴ αὐτὴ τῆς ὁμοιότητος, ἀναγκαῖον δὲ ᾖ καὶ τῷ φορτικῷ τούτῳ προσχρῆσθαι, τῇ ξυνθήκῃ, εἰς ὀνομάτων ὀρθότητα· ἐπεὶ ἴσως κατά γε τὸ δυνατὸν κάλλιστ’ ἂν λέγοιτο, ὅταν ἢ πᾶσιν ἢ ὡς πλείστοις ὁμοίοις λέγηται, τοῦτο δ’ ἐστὶ προσήκουσιν, αἴσχιστα δὲ τοὐναντίον.
There are names better — more like, or less like to the things named: Natural Names are the best, but they cannot always be had. Names may be significant by habit, though in an inferior way.
You say (reasons Sokrates with Kratylus) that names must be significant by way of likeness. But there are degrees of likeness. A portrait is more or less like its original, but it is never exactly like: it is never a duplicate, nor does it need to be so. Or a portrait, which really belongs to and resembles one person, may be erroneously assigned to another. The same thing happens with names. There are names more or less like the thing named — good or bad: there are names good with reference to their own object, but erroneously fitted on to objects not their own. The name does not cease to be a name, so long as the type or form of the thing named is preserved in it: but it is worse or better, according as the accompanying features are more or less in harmony with the form.[69] If names are like things, the letters which are put together to form names, must have a natural resemblance to things — as we remarked above respecting the letters Rho, Lambda, &c. But the natural, inherent, powers of resemblance and significance, which we pronounced to belong to these letters, are not found to pervade all the actual names, in which they are employed. There are words containing the letters Rho and Lambda, in a sense opposite to that which is natural to them — yet nevertheless at the same time significant; as is evident from the fact, that you and I and others understand them alike. Here then are words significant, without resembling: significant altogether through habit and convention. We must admit the principle of convention as an inferior ground and manner of significance. Resemblance, though the best ground as far as it can be had, is not the only one.[70]
[69] Plato, Kratyl. pp. 432-434.